CN115051930A - Defect optimization method based on AISeOps combined with middleware algorithm - Google Patents
Defect optimization method based on AISeOps combined with middleware algorithm Download PDFInfo
- Publication number
- CN115051930A CN115051930A CN202210560064.1A CN202210560064A CN115051930A CN 115051930 A CN115051930 A CN 115051930A CN 202210560064 A CN202210560064 A CN 202210560064A CN 115051930 A CN115051930 A CN 115051930A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- defect
- algorithm model
- data
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/04—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks
- H04L63/0428—Network architectures or network communication protocols for network security for providing a confidential data exchange among entities communicating through data packet networks wherein the data content is protected, e.g. by encrypting or encapsulating the payload
Abstract
The defect optimization method based on AISeOps combined with the middle stage algorithm comprises the following steps of S1: the data acquisition module acquires and stores data information in the distributed network and then transmits the data information to the AIOps algorithm module; s2: the AIOps algorithm module analyzes the algorithm model applied by each service in operation and maintenance work according to the data information to further form a plurality of algorithm model sets, so that each service in operation and maintenance work is matched with the corresponding algorithm model sets; and analyzing the algorithm model of the business application by the algorithm defect capturing unit regularly, and optimizing by the algorithm defect optimizing unit when the algorithm model has defects in operation. According to the method and the device, various algorithms applied to the operation and maintenance work are subjected to defect capture, and the algorithms applied to the work are optimized, so that the algorithm model used in the operation and maintenance work is more consistent with the service scene, and a more complete automatic operation and maintenance effect is realized.
Description
Technical Field
The invention relates to the technical field of network operation and maintenance, in particular to a defect optimization method based on AISeOps combined with a middleman algorithm.
Background
With the large-scale application of the artificial intelligence technology, the problems of labor cost and efficiency of repetitive operation and maintenance work in the traditional automatic operation and maintenance system are effectively solved. However, in the process of fault handling, change management, capacity management, and service resources in a complex scenario, a person is still required to control the decision process, which hinders further improvement of operation and maintenance efficiency. The operation and maintenance data center is used as a drive, an AI algorithm is used as a core, professional operation and maintenance service modules such as infrastructure monitoring, fault accurate positioning and intelligent processing, 3D digital twin, management cockpit and the like are covered, production and research are combined, the requirement of the whole industry can be met, and a stable, reliable, complete-function, advanced-technology and autonomous-controllable full-stack intelligent operation and maintenance platform is created. Thereby making it possible to realize automatic operation and maintenance in real sense.
The technical scheme provided by the application is mainly based on AISeOps combined with a middlebox algorithm, carries out defect capture on various algorithms applied to the operation and maintenance work business, and optimizes the algorithms applied to the business, so that the algorithm models used in the operation and maintenance work are more consistent with the business scene, and a more complete automatic operation and maintenance effect is realized.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a defect optimization method based on AISeOps combined with a middlebox algorithm, introduces an intelligent operation and maintenance system based on AISeOps module chain combined middlebox (the system takes an operation and maintenance data middlebox as a drive and an AI algorithm as a core), captures the defects of various algorithms applied in the operation and maintenance work business, and optimizes the algorithm applied by the business, so that the algorithm model used in the operation and maintenance work is more consistent with the business scene, and the more complete automatic operation and maintenance effect is realized.
In order to achieve the purpose, the invention adopts the following technical scheme:
the defect optimization method based on AISeOps combined with the middleware algorithm is characterized by comprising the following steps of:
s1: the data acquisition module acquires and stores data information in the distributed network and then transmits the data information to the AIOps algorithm module;
s2: the AIOps algorithm module analyzes the algorithm model applied by each service in operation and maintenance work according to the data information to further form a plurality of algorithm model sets, so that each service in operation and maintenance work is matched with the corresponding algorithm model sets;
and analyzing the algorithm model of the business application by the algorithm defect capturing unit regularly, and optimizing by the algorithm defect optimizing unit when the algorithm model has defects in operation.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, the data information content in step S1 includes the original data of the distributed network, and the privacy data in the distributed network, where the privacy data is the data whose state is legal after being verified by the middlebox based on the module chain digital identity certificate.
Further, the specific content of step S2 is:
s2.1: the AIOps algorithm module receives the data information of the data acquisition module, analyzes an algorithm model applied to each service in operation and maintenance work according to the data information, and matches and combines each service and the corresponding algorithm model to obtain an algorithm model set, so that the services correspond to the algorithm model set one by one; each algorithm model set is provided with a corresponding initial algorithm set identifier; the format of the initial algorithm set identifier is: a data acquisition module ID + an algorithm model ID;
s2.2: the algorithm defect capturing unit works regularly, each algorithm model in the algorithm model set is trained according to business requirements in order according to real historical data of the algorithm model applied in business, whether defects exist in each algorithm model is analyzed through a training result, if the training result is normal, the algorithm defect state identification of the algorithm model set where the algorithm model is located is 0, and if the training result is abnormal, the algorithm defect state identification of the algorithm model set where the algorithm model is located is 1;
the algorithm defect capturing unit sends the algorithm defect state identification to an algorithm defect database for storage; the algorithm cheat end database is also used for storing the names and IDs of all algorithm models, the names and IDs of associated algorithm models corresponding to all algorithm models, and the names and IDs of all algorithm models and the optimized algorithm models corresponding to the associated algorithm models;
s2.3: aiming at algorithm models with defects, an algorithm defect optimizing unit takes the ID of the algorithm model and the associated ID of the algorithm model as retrieval conditions, and accesses an algorithm defect database to obtain the ID of the optimized algorithm model under the corresponding retrieval conditions; meanwhile, inquiring and comparing other algorithm model IDs in the algorithm model set with the optimization algorithm ID; if the comparison results are different, the optimization algorithm ID is put into the algorithm model set of the service, and if the comparison results are the same, the optimization algorithm ID is not put into the algorithm model set of the service; updating the initial algorithm set identification is completed, the optimized algorithm set identification is obtained, and the optimization of the algorithm model set of the corresponding service is achieved;
the specific format of the optimized algorithm set identifier is as follows: the method comprises the following steps that (1) a data acquisition module ID + algorithm model ID + optimization algorithm model ID + algorithm defect state identification; and the state identifier of the algorithm defect in the optimized algorithm model set identifier is updated to 0 from 1.
Further, the method also includes step S3: and encrypting the optimized algorithm set identification by adopting an SHA-256 digital encryption mode, and judging whether the encrypted data is tampered or not through verification.
Further, the specific content of step S3 is:
bit filling: taking binary data information identified by the optimized algorithm set as source information, performing bit complementing on the tail end of the source information by one 1, and then performing bit complementing on a string of 0 after 1, so that the remainder is 448 after modulus of the source information after bit complementing is performed on 512;
length supplement: the binary data length corresponding to the source information is put behind the data which is subjected to bit complementing;
partitioning: if the binary length corresponding to the source information after bit complementing exceeds 512, or the binary length corresponding to the source information after the bit complementing exceeds a multiple of 512, partitioning according to 512 bits; otherwise, converting the binary data corresponding to the source information after the length is supplemented into hexadecimal to carry out the existing length partitioning;
and (3) forming an encryption string: dividing each block into 8 64-bit binary systems, extracting 32bits of each binary system, and connecting the binary systems to obtain a Hash value digital encryption string with the length of 256 bits;
and (3) verification: and intercepting the first four digits of the digital encryption string, putting the intercepted first four digits into a middle station digital encryption database for verification, finishing strong countermeasure verification of the digital encryption string and generating an effective digital encryption string if the intercepted first four digits are not the same in repetition, and storing the effective digital encryption string in the middle station digital encryption database.
Further, the method also includes step S4: and synchronizing the processing contents in the steps S1-S3 into a Docker container containing the application module in each network node in the distributed network.
The beneficial effects of the invention are: the invention adopts the digital encryption technology to manage the algorithm set identification, introduces the design concept of the operation and maintenance data middle station, serially connects all levels of the middle station in a module mode, captures and analyzes the defects of the algorithm applied by the service, and generates the optimized algorithm set identification by the algorithm defect optimizing unit, thereby enabling the algorithm of the operation and maintenance service to be more accurately and completely used. And the Docker container solves the problem of the consistency of the algorithm model and the data, provides a foundation for further mining the data, and well adapts to the diversity requirement of foreground application. The intelligent operation and maintenance intermediate station is creatively combined with the digital encryption technology, so that a method for penetrating AISeOps intelligent security operation technology to the AIOps intermediate station is provided.
Detailed Description
The present invention will now be described in further detail.
This application mainly includes: and the data center function module, the privacy data module chain digital encryption verification and the Docker consistency data are synchronized to the application module for display and application modules.
Step one, the data center function module mainly comprises: the system comprises a data acquisition and analysis module, an AIOps algorithm module and a CI configuration library module.
1-1 data acquisition and analysis module:
firstly, the AIOps algorithm module extracts key information from the distributed network original data of the data acquisition and analysis module and utilizes the key information, and private data is extracted and provided for the AIOps algorithm module through the data of which the state is legal after the data is digitally hashed and encrypted by the middle station.
The specific description is as follows: storing original data of the distributed network and private data of each part needing to participate in operation. The data source includes: environment monitoring, network monitoring, host monitoring, system monitoring, security monitoring, cloud resource monitoring and application monitoring. Key information is extracted from the data source collected distributed network original data and utilized, and private data is extracted from the data of the middle station based on the state legality after module digital Hash encryption and provided to the AIOps algorithm module for model operation and data cleaning and filtering.
1-2AIOps Algorithm:
secondly, the AIOps algorithm module accesses a common machine learning algorithm and forms an AIOps algorithm set according to the specific scenes of operation and maintenance work. The algorithm defect capturing program and the algorithm defect optimizing program which are creatively added are used for complementarily calling the optimization algorithm with the current algorithm defect, so that the problems that the current algorithm is distorted in operation, cannot be continuously operated and the like are solved.
Table 1 introduction of algorithm defect database optimization structure:
the markov chain ID is 1 ant colony algorithm ID 2 bayesian algorithm ID 3 pareto analysis ID 4.
1-3CI configuration library module:
and then, the CI configuration library module is used as an initial reference of the intelligent operation and maintenance system, and unified standardized management is carried out on all components in the intelligent operation and maintenance system aiming at special scenes in operation and maintenance work. Namely, when a scene needs to be created, the data acquisition and analysis module, the AIOps algorithm module and the application module are connected in series and managed in a unified mode through the CI configuration library configuration.
The specific description is as follows: the CI configuration library module is used as an initial reference of the intelligent operation and maintenance system, and unified standardized management is carried out on all components in the intelligent operation and maintenance system aiming at special scenes in operation and maintenance work. The sub-modules include: CI configuration item management, CI relation management, service model topology, dictionary and rule management.
And step two, adopting SHA-256 digital encryption verification by the middleboxes (algorithm set identifications). The data is prevented from being tampered in the transmission process of the AIOps algorithm module through verification.
Initial [ algorithm set identification ] format: the data acquisition module ID + algorithm module ID includes (algorithm model 1+ ·+ algorithm model N).
After optimization, [ algorithm set identification ] format: the method comprises the steps of data acquisition module ID + algorithm module ID + (algorithm Model 1+ optimization algorithm Model 2+ ·+ algorithm Model N) + algorithm defect state identification (Model ID).
Step three, synchronizing Docker consistency data to application module for display
And transmitting the operation result and the data to a Docker container containing the application module on each network node in the network group through a database or an API (application programming interface), thereby maintaining the display consistency of the application module.
The following description will be made with reference to specific examples (markov chain model, ant colony algorithm model, optimization of adverse effects).
This application mainly includes: and the data center function module, the privacy data module chain digital encryption verification and the Docker consistency data are synchronized to the application module for display and application modules.
Step one, the data center function module mainly comprises: the system comprises a data acquisition and analysis module, an AIOps algorithm module and a CI configuration library module.
1-1 data acquisition module:
firstly, the AIOps algorithm module extracts key information from the distributed network original data of the data acquisition and analysis module and utilizes the key information, and private data is extracted and provided for the AIOps algorithm module through data with a legal state verified by a middle station based on a module chain digital identity certificate.
The specific description is as follows: storing original data of the distributed network and private data of each part needing to participate in operation. The data source includes: environment monitoring, network monitoring, host monitoring, system monitoring, security monitoring, cloud resource monitoring and application monitoring. And key information is extracted from the data source collected distributed network original data and is utilized, and the private data is extracted from the data with a legal state verified by the middle station based on the module chain digital identity certificate and is provided for the AIOps algorithm module for model operation and data cleaning and filtering.
1-2AIOps Algorithm:
secondly, the AIOps algorithm module accesses a common machine learning algorithm and forms an AIOps algorithm set according to the specific scenes of operation and maintenance work. The method is characterized in that an algorithm defect capturing program and an algorithm defect optimizing program are creatively added, and the optimization algorithm with the current algorithm defect is obtained for complementary calling, so that the problems that the operation of the current algorithm is distorted, and the learning result of the algorithm is repeated circularly to have the same result due to the learning data problem, so that the continuous learning is not realized, the calculated data is abnormal and the like are overcome.
1-2-S1, firstly, after receiving legal state data, selecting algorithm models required by the service for combination. And simultaneously generating an initial [ algorithm set identifier ] format: the data acquisition module ID + algorithm module ID includes (algorithm model 1+ ·+ algorithm model N).
1-2-S2, secondly, executing an algorithm defect capturing program, wherein a catcher belongs to a preset script, executing the script according to time intervals, collecting real historical data trained by each algorithm Model through simulating real services, orderly putting the real historical data into the Model according to service requirements for training, and analyzing a training result to obtain an algorithm defect state identifier (Model ID). The state is 0 normal, and 1 is abnormal. And storing the algorithm defect state identification and the algorithm model name into an algorithm defect database. And (4) setting the algorithm defect state identifier (Model ID) to be 1, and sending the identifier to an algorithm defect optimization program.
Detailed description of algorithm drawbacks 1: in the nth time period, the fault and non-fault occurrence probability of the operation node is converted into stable distribution (0.510.49) finally along with the continuation, the same probability is obtained no matter how many times the operation node is calculated, parameters of the Model need to be changed at this time, but for the client parameters of the stable service, the Model is continuously operated by executing an algorithm defect capturing program to know that the occurrence probability is the same probability, and then a defect state identifier (Model ID) of the Markov chain Model in an algorithm defect database is updated to be set to be 1.
[ Markov chain model ] the results of the drawbacks of the operation describe:
performing an algorithm complementation model on the operation result: the Bayes + Markov chain can make up the problem that the Markov chain can not continuously operate. And optimizing the probability analysis result.
Markov chain algorithm operation defect flow formula: x (k +1) ═ X (k) X P
In the formula: x (k) represents a state vector of the trend analysis and prediction target at time t ═ k, P represents a one-step transition probability matrix, and X (k +1) represents a state vector of the trend analysis and prediction target at time t ═ k + 1. And generating a data set by adopting a two-step transfer matrix.
Examples are: (three sets of data required for Markov chain model)
Initial probability of failure of historical network node (0.3, 0.7)
Probability of network node fault transfer to non-fault in current period [ 0.6, 0.4 ]
Probability of normal transfer of network node to fault [ 0.3, 0.7 ] in current period
The ratio of the future failures of the network nodes in the network group is obtained through the first operation
Lower period node failure occurrence probability 0.3x0.6+0.3x 0.7-0.39
Lower interval node normal occurrence probability 0.3x0.4+0.7x0.7 ═ 0.61
Probability of node failure and non-failure occurrence in lower period [ 0.390.61 ]
The initial probability (0.3, 0.7) is changed into (0.390.61) for the second time
Lower interval node fault occurrence probability 0.39x0.6+0.61x0.3 ═ 0.417
Lower interval node normal occurrence probability 0.39x0.4+0.61x0.7 ═ 0.583
Probability of node failure and non-failure occurrence in lower period [ 0.4170.583 ]
The result of the Nth operation is smoothly distributed [ 0.490.51 ]
The N +1 th operation result is distributed smoothly (0.490.51)
And after the Model is continuously operated by executing an algorithm defect capturing program to know that the occurrence probabilities are the same, updating a defect state identifier (Model ID) of the Markov chain algorithm Model in an algorithm defect database to be 1.
Markov chain algorithm model intelligent operation and maintenance application scene: fault early warning, network security, remote disaster recovery, twin network, automatic discovery, root cause analysis, early warning baseline and the like.
The algorithm disadvantages are specifically described 2:
the core formula and the description are as follows:
And (3) executing an algorithm defect capturing program to calculate the ant colony algorithm Model, wherein if the obtained result is not unique and a fastest path from A to B needs to be provided in the networking, the calculation result gives a plurality of lines to indicate that the algorithm has defects, and the defect state identifier (Model ID) of the ant colony algorithm Model in the algorithm defect database is set to be 1.
1-2-S3, constructing an algorithm defect optimization program:
specifically describing an algorithm defect optimization program:
firstly, accessing an algorithm defect database according to the transmitted algorithm model name and the associated algorithm ID as retrieval conditions to obtain an optimal optimization algorithm ID corresponding to the algorithm defect of the current algorithm model, which is abbreviated as: the optimal optimization algorithm ID. A plurality is comma spaced.
And secondly, comparing the algorithm IDs of other combinations in the transmitted algorithm set with the optimal optimization algorithm ID, if the comparison results are different, putting the optimal optimization algorithm ID into the algorithm set, and if the comparison results are the same, not putting the optimal optimization algorithm ID into the algorithm set. And finishing the operation of the algorithm defect optimization model. And obtaining an optimized algorithm set, thereby overcoming the problems of algorithm operation distortion, incapability of continuous operation and the like.
Example one: one of the best optimization schemes for the disadvantages of Markov chain models
When the Nth operation result appears [ 0.490.51 ]
The result of the (N +1) th operation still appears [ 0.490.51 ]
Three groups of rectangular data for representing the model in a stable distribution state need to be adjusted:
probability of network node fault transfer to non-fault in current time period [ 0.6, 0.4 ]
Probability of normal transfer of network node to fault [ 0.3, 0.7 ] in current period
And correcting the abnormal value by using Bayes prior probability and a Bayes model according to the abnormal value and the Markov transfer matrix to obtain the corrected recognition probability. The Markov chain model can be continuously operated, the influence of stable distribution on model operation is solved, and meanwhile, the optimal defect optimization model ID in the algorithm defect database is updated to be set as the ID of the Bayesian model. Thereby completing the optimization of the Markov chain model algorithm.
Example two: one of the best optimization schemes for the disadvantages of the ant colony algorithm
The ant colony algorithm obtains the shortest route in a plurality of routes, the operation result of the drawback has a plurality of routes, and the unique result cannot be obtained through optimization. Therefore, Pareto Analysis, also called ABC classification, is adopted as Pareto Analysis, also called principal and secondary factor Analysis. And when a non-unique optimal result is obtained, activating a pareto analysis method, and optimizing the same optimal result again to obtain a process of obtaining another optimal result while not influencing the efficiency of one result.
When the ant colony results have non-unique results and only unique operation results are needed for business requirements, multiple intelligent model algorithms are usually adopted to optimize the operation results and obtain the unique operation. The ant colony algorithm operation result is optimized by using a pareto analysis method.
Pareto analytical formula: XA + XB ═ X1; YA + YB ═ Y1
And respectively putting the data of the non-unique operation results of the ant colony algorithm into a pareto analysis method formula to obtain the optimal probability of each result, wherein the higher the probability is, the closer the result is to the optimal result from high to low. And meanwhile, updating the ID of the optimal defect optimization model in the algorithm defect database to be the ID of the model of the pareto analysis method. Thereby completing the optimization of the ant colony algorithm model algorithm defects. The ant colony algorithm model intelligent operation and maintenance application scene: fault self-healing, intelligent traffic scheduling, fault tracing, intelligent network attack and defense drilling and the like.
1-2-S4, then, obtaining an optimized algorithm set, updating an algorithm defect state identifier (Model ID) ═ 1 to 0, and regenerating necessary parameters such as an algorithm Model name, a module ID, a related algorithm ID and the like into optimized [ algorithm set identifier ].
After optimization, [ algorithm set identification ] format: the method comprises the steps of data acquisition module ID + algorithm module ID + (algorithm Model 1+ optimization algorithm Model 2+ ·+ algorithm Model N) + algorithm defect state identification (Model ID). And meanwhile, updating the data acquisition module ID of the database with the algorithm defects, the algorithm module ID (comprising the algorithm model 1+ the optimization algorithm model 2+.. + the associated algorithm model N) and the algorithm defect state identification, and finally submitting to the privacy data module chain digital encryption verification in the second step.
1-3CI configuration library module:
and then, the CI configuration library module is used as an initial reference of the intelligent operation and maintenance system, and unified standardized management is carried out on all components in the intelligent operation and maintenance system aiming at special scenes in operation and maintenance work. Namely, when a scene needs to be created, the data acquisition and analysis module, the AIOps algorithm module and the application module are connected in series and managed in a unified mode through the CI configuration library configuration.
The specific description is as follows: and the CI configuration library module is used as an initial reference of the intelligent operation and maintenance system, and is used for performing unified standardized management on all components in the intelligent operation and maintenance system aiming at special scenes in operation and maintenance work. The sub-modules include: CI configuration item management, CI relation management, service model topology, dictionary and rule management.
And step two, adopting SHA-256 digital encryption by the middleboxes (algorithm set identifiers). The data is prevented from being tampered in the transmission process of the AIOps algorithm module through verification.
Initial [ algorithm set identifier ] format: the data acquisition module ID + algorithm module ID includes (algorithm model 1+ ·+ algorithm model N).
After optimization, [ algorithm set identification ] format: the method comprises the steps of data acquisition module ID + algorithm module ID + (algorithm Model 1+ optimization algorithm Model 2+ ·+ algorithm Model N) + algorithm defect state identification (Model ID).
The specific description is as follows: and (4) carrying out bit supplementing and length supplementing on the plaintext (algorithm set identifier).
First, SHA-256 must complement the source data. The goal is to make the remainder 448 after modulo 512 its length. The first step of bit padding is to first pad one 1 in the last bit. The second step is to complement a string of 0's later so that the length of the padded data satisfies the requirement that the remainder is 448 after modulo 512.
Secondly, length supplementing and blocking operation: the length of binary data corresponding to the original data is put after the data that has been subjected to the complementary bit. And converted to hexadecimal. And if the binary length of the original data exceeds 512 and the length-complemented data exceeds the multiple of 512, partitioning according to 512 bits. No more than 512 blocks per existing length.
Then, each 512bits is divided into 8 64-bit binaries and each binary 32-bit is extracted, concatenated, i.e., 256-bit long Hash value digital encryption string. And intercepting the first four digits of the digital encryption string, putting the intercepted first four digits into a digital encryption database of the middle station for verification, finishing strong countermeasure verification of the digital encryption string if the intercepted first four digits are not repeated, generating an effective digital encryption string, and storing the effective digital encryption string in the digital encryption database of the middle station.
Step three, synchronizing Docker consistency data to application module for display
And transmitting the operation result and the data to each network node in the network group through a database or an API (application programming interface) in a Docker container containing the application module, thereby maintaining the display consistency of the application module.
3-1 application module:
the specific description is as follows: the application layer has the main advantages of providing better support for further mining of the foreground data and meeting the diversity requirement of foreground application. The sub-modules include: the system comprises a large screen, a cab, a fault accurate positioning device, a fault automatic processing device, a fault prediction device, a digital twin device, an intelligent alarm analysis device, an intelligent order dispatching device, an intelligent report and a mobile office.
What is needed is: the application creatively combines the intelligent operation and maintenance intermediate station with the Hash digital encryption technology, thereby penetrating the AISeOps intelligent security operation technology to the AIOps intermediate station. Firstly, when a service scene needs to be created, the data acquisition and analysis module, the AIOps algorithm module and the application module are connected in series and managed in a unified mode through the CI configuration library configuration. Secondly, the AIOps algorithm module stores common machine learning algorithms and models, and an AIOps algorithm set is formed according to the specific scenes of operation and maintenance work. And thirdly, the AIOps algorithm module extracts and utilizes key information from the distributed network original data set of the data acquisition and analysis module, generates an initial algorithm set identifier and adopts Hash digital encryption to prevent the data from being tampered in the module transmission process. Then, the data is extracted and provided to the AIOps algorithm module. The AIOps algorithm module creatively adds an algorithm defect capturing program and an algorithm defect optimizing program. The algorithm defect capturing program puts real historical data into a Model for training by simulating real services, generates an algorithm defect state identifier (Model ID) with the exception of 1 according to a training result, and stores the algorithm defect state identifier (Model ID) with the exception of 0 into an algorithm defect database. Meanwhile, an optimal optimization algorithm ID is obtained by executing an algorithm defect optimization procedure, is added into the current algorithm set, and is updated (Model ID) to be 0 normally, so that the problems of operation distortion, incapability of continuous operation and the like of the current algorithm are solved. And finally, transmitting the optimized [ algorithm set identification ] to each network node in the group network through a Docker container, and keeping the display consistency of the application modules.
It should be noted that the terms "upper", "lower", "left", "right", "front", "back", etc. used in the present invention are for clarity of description only, and are not intended to limit the scope of the present invention, and the relative relationship between the terms and the terms is not limited by the technical contents of the essential changes.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (6)
1. The defect optimization method based on AISeOps combined with the middling station algorithm is characterized by comprising the following steps of:
s1: the data acquisition module acquires and stores data information in the distributed network and then transmits the data information to the AIOps algorithm module;
s2: the AIOps algorithm module analyzes the algorithm model applied by each service in operation and maintenance work according to the data information to further form a plurality of algorithm model sets, so that each service in operation and maintenance work is matched with the corresponding algorithm model set;
and analyzing the algorithm model of the business application by the algorithm defect capturing unit regularly, and optimizing by the algorithm defect optimizing unit when the algorithm model has defects in operation.
2. The AISeOps-based defect optimization method combined with the SSO algorithm as claimed in claim 1, wherein the data information content in step S1 includes original data of the distributed network and private data in the distributed network, wherein the private data is the data whose state is legal after being verified by the SSO based on the module chain digital identity certificate.
3. The AISeOps combination middling station algorithm-based defect optimization method of claim 1, wherein the specific content of step S2 is as follows:
s2.1: the AIOps algorithm module receives data information of the data acquisition module, analyzes an algorithm model applied to each service in operation and maintenance work according to the data information, and matches and combines each service and the corresponding algorithm model to obtain an algorithm model set, so that the services correspond to the algorithm model set one by one; each algorithm model set is provided with a corresponding initial algorithm set identifier; the format of the initial algorithm set identifier is: a data acquisition module ID + an algorithm model ID;
s2.2: the algorithm defect capturing unit works regularly, trains each algorithm model in the algorithm model set according to the business requirements in order according to the real historical data of the algorithm model applied in the business, analyzes whether each algorithm model has defects or not according to the training result, if the training result is normal, the algorithm defect state identification of the algorithm model set where the algorithm model is located is 0, and if the result is abnormal, the algorithm defect state identification of the algorithm model set where the algorithm model is located is 1;
the algorithm defect capturing unit sends the algorithm defect state identification to an algorithm defect database for storage; the algorithm defect database is also used for storing the names and IDs of all algorithm models, the names and IDs of the associated algorithm models corresponding to all algorithm models, and the names and IDs of all algorithm models and the optimized algorithm models corresponding to the associated algorithm models;
s2.3: aiming at algorithm models with defects, an algorithm defect optimizing unit takes the algorithm model ID and the associated algorithm model ID as retrieval conditions, and accesses an algorithm defect database to obtain an optimization algorithm model ID under the corresponding retrieval conditions; meanwhile, inquiring and comparing other algorithm model IDs in the algorithm model set with the optimization algorithm ID; if the comparison results are different, the optimization algorithm ID is put into the algorithm model set of the service, and if the comparison results are the same, the optimization algorithm ID is not put into the algorithm model set of the service; updating the initial algorithm set identification is completed in this way, and the optimized algorithm set identification is obtained, so that the optimization of the algorithm model set of the corresponding service is achieved;
the specific format of the optimized algorithm set identifier is as follows: the method comprises the following steps that (1) a data acquisition module ID + algorithm model ID + optimization algorithm model ID + algorithm defect state identification; and the state identifier of the algorithm defect in the optimized algorithm model set identifier is updated to 0 from 1.
4. The AISecOps combination Zhongtai algorithm-based defect optimization method of claim 3, further comprising the step S3 of: and encrypting the optimized algorithm set identification by adopting an SHA-256 digital encryption mode, and judging whether the encrypted data is tampered by verification.
5. The AISeOps combination Zhongtai algorithm-based defect optimization method of claim 4, wherein the step S3 includes the following steps:
bit filling: taking binary data information identified by the optimized algorithm set as source information, performing bit complementing on the tail end of the source information by one 1, and then performing bit complementing on a string of 0 after 1, so that the remainder is 448 after modulus of the source information after bit complementing is performed on 512;
length supplement: the binary data length corresponding to the source information is put behind the data which is subjected to bit complementing;
partitioning: if the binary length corresponding to the source information after bit complementing exceeds 512, or the binary length corresponding to the source information after the bit complementing exceeds a multiple of 512, partitioning according to 512 bits; otherwise, converting the binary data corresponding to the source information after the length is supplemented into hexadecimal to carry out the existing length partitioning;
and (3) forming an encryption string: dividing each block into 8 64-bit binary systems, extracting 32bits of each binary system, and connecting to obtain a Hash value digital encryption string with the length of 256 bits;
and (3) verification: and intercepting the first four digits of the digital encryption string, putting the intercepted first four digits into a digital encryption database of the middle station for verification, finishing strong countermeasure verification of the digital encryption string if the intercepted first four digits are not the same as the intercepted first four digits, generating an effective digital encryption string, and storing the effective digital encryption string in the digital encryption database of the middle station.
6. The AISecOps combination Zhongtai algorithm-based defect optimization method of claim 4, further comprising the step S4 of: and synchronizing the processing contents in the steps S1-S3 into a Docker container containing the application module in each network node in the distributed network.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210560064.1A CN115051930B (en) | 2022-05-23 | 2022-05-23 | AISecOps combined middle station algorithm-based defect optimization method |
| PCT/CN2022/114209 WO2023226215A1 (en) | 2022-05-23 | 2022-08-23 | Defect optimization method based on aisecops in combination with middle platform algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210560064.1A CN115051930B (en) | 2022-05-23 | 2022-05-23 | AISecOps combined middle station algorithm-based defect optimization method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115051930A true CN115051930A (en) | 2022-09-13 |
| CN115051930B CN115051930B (en) | 2023-05-12 |
Family
ID=83159074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210560064.1A Active CN115051930B (en) | 2022-05-23 | 2022-05-23 | AISecOps combined middle station algorithm-based defect optimization method |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115051930B (en) |
| WO (1) | WO2023226215A1 (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112181960A (en) * | 2020-09-18 | 2021-01-05 | 杭州优云软件有限公司 | Intelligent operation and maintenance framework system based on AIOps |
| CN112182077A (en) * | 2020-09-11 | 2021-01-05 | 杭州优云软件有限公司 | Intelligent operation and maintenance system based on data middling platform technology |
| US20210271582A1 (en) * | 2018-06-28 | 2021-09-02 | Zte Corporation | Operation and maintenance system and method |
| US20210295158A1 (en) * | 2020-03-17 | 2021-09-23 | Onspecta, Inc. | End-to-end optimization |
| CN113516244A (en) * | 2021-07-27 | 2021-10-19 | 盛景智能科技(嘉兴)有限公司 | Intelligent operation and maintenance method and device, electronic equipment and storage medium |
| CN114139747A (en) * | 2021-12-09 | 2022-03-04 | 国网河北省电力有限公司信息通信分公司 | AIOps intelligent operation and maintenance system based on artificial intelligence technology |
| CN114139949A (en) * | 2021-12-01 | 2022-03-04 | 成都西交轨道交通技术服务有限公司 | Intelligent operation and maintenance system and method for rail transit based on edge calculation and machine learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11165707B2 (en) * | 2019-04-12 | 2021-11-02 | Cisco Technology, Inc. | Dynamic policy implementation for application-aware routing based on granular business insights |
| CN114244687B (en) * | 2021-12-20 | 2023-08-08 | 中电信数智科技有限公司 | Network fault self-healing operability judging method based on AIOps |
-
2022
- 2022-05-23 CN CN202210560064.1A patent/CN115051930B/en active Active
- 2022-08-23 WO PCT/CN2022/114209 patent/WO2023226215A1/en unknown
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210271582A1 (en) * | 2018-06-28 | 2021-09-02 | Zte Corporation | Operation and maintenance system and method |
| US20210295158A1 (en) * | 2020-03-17 | 2021-09-23 | Onspecta, Inc. | End-to-end optimization |
| CN112182077A (en) * | 2020-09-11 | 2021-01-05 | 杭州优云软件有限公司 | Intelligent operation and maintenance system based on data middling platform technology |
| CN112181960A (en) * | 2020-09-18 | 2021-01-05 | 杭州优云软件有限公司 | Intelligent operation and maintenance framework system based on AIOps |
| CN113516244A (en) * | 2021-07-27 | 2021-10-19 | 盛景智能科技(嘉兴)有限公司 | Intelligent operation and maintenance method and device, electronic equipment and storage medium |
| CN114139949A (en) * | 2021-12-01 | 2022-03-04 | 成都西交轨道交通技术服务有限公司 | Intelligent operation and maintenance system and method for rail transit based on edge calculation and machine learning |
| CN114139747A (en) * | 2021-12-09 | 2022-03-04 | 国网河北省电力有限公司信息通信分公司 | AIOps intelligent operation and maintenance system based on artificial intelligence technology |
Non-Patent Citations (2)
| Title |
|---|
| ANNA LEVIN: "AIOps for a Cloud Object Storage Service", 2019 IEEE INTERNATIONAL CONGRESS ON BIG DATA * |
| 杜永生;: "智能运维,基于自学习的自动化运维", 信息通信技术 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023226215A1 (en) | 2023-11-30 |
| CN115051930B (en) | 2023-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106100999B (en) | Image network flow control methods in a kind of virtualized network environment | |
| CN110298381B (en) | Cloud security service function tree network intrusion detection system | |
| CN106416132A (en) | Systems and methods for controlling network switches using a switch modeling interface at a controller | |
| CN110009201B (en) | Electric power data link system and method based on block chain technology | |
| CN114389940A (en) | Failure recovery plan determination method, device and system and computer storage medium | |
| CN105376246A (en) | Adaptive generation management system and method of security strategy based on SDN | |
| CN101640825A (en) | Integration of three networks | |
| CN108234315A (en) | Image network flow control protocol in a kind of virtualized network environment | |
| CN109714206A (en) | Electric power monitoring system Generating Network Topology Map, network bus topological diagram | |
| CN106603550A (en) | Network isolation method and network isolation device | |
| CN106453541A (en) | Data synchronization method, server and data synchronization system | |
| CN112929200B (en) | SDN multi-controller oriented anomaly detection method | |
| CN106254457A (en) | Real estate data-sharing systems based on WCF technology and data sharing method | |
| CN106203164B (en) | Information security big data resource management system based on trust computing and cloud computing | |
| CN114415732B (en) | Unmanned aerial vehicle bee colony ground command control system based on multistage formation | |
| CN112468592A (en) | Terminal online state detection method and system based on electric power information acquisition | |
| CN116366673A (en) | Data analysis and sharing method oriented to management of power grid information machine room | |
| CN114124956A (en) | Block chain edge calculation system | |
| CN113553374A (en) | Decentralized distributed large data block chain information storage system | |
| CN115051930A (en) | Defect optimization method based on AISeOps combined with middleware algorithm | |
| CN111476656B (en) | Transaction safety identification method based on block chain | |
| CN106603722A (en) | Management device determining method and device | |
| CN103401791A (en) | Method and equipment for identifying boundary port | |
| CN106302006A (en) | A kind of dynamic source tracing method of IP spoofing packet based on SDN | |
| CN108989475A (en) | Distributed storage transmission architecture based on block chain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |